The VASARI project utilizes a specialized algorithm for stitching together the sub-images of its subject painting. This is made possible because the system's exact registration allows for point to point matching. Again, our registration was not perfect, so we resorted to commercial image mosaicing software.
 

            We purchased QuickStitch 2.0, published by Enroute Imaging, a company based in Palo Alto. This software loads up to 48 sub-images and stitches them together into one large high-resolution image. Since it is consumer-oriented software, there were only a few options to control the stitching algorithm. These were the choices between perspective and panhead mapping, and between gradual and sharp blending. Through correspondence with the engineering manager of Enroute Imaging we learned more about these options:
 

"The Sharp/Gradual distinction dictates whether overlapping images are cross-blended or not. Cross-blending results in each image being partially visible in a semi-transparent way; when the images are not perfectly aligned, this results in a "ghosting" effect where slightly offset, semi-transparent duplicates of elements of the scene are visible. The Sharp setting turns off gradual blending, which makes the transition from one picture to another an abrupt one at some arbitrarily chosen seam.  The benefit of Gradual is that it deals better with adjacent images that have different lighting characteristics.  The benefit of Sharp is that it can minimize ghosting.

The distinction between Perspective and Panhead is the distinction between seeing the composite image projected onto a flat wall vs. seeing it projected onto a cylinder that wraps around your point of view."    -(from email correspondence)
 
 

            We chose the image rendered with the Panhead algorithm and Sharp blending as best.